Rapid Survey on Routing in Data Centers

Rapid Survey on Routing in Data Centers

Rapid Survey on Routing in Data Centers

Ghazal Tashakor

Data Center Structures

Server-Centric

Switch-Centric

Hybrid Electrical/Optical

Routing in Server-Centric

Bcube

DCell

Routing in Switch-Centric

PortLand

VL2

Routing in Hybrid Electrical/Optical

C-Through

Helios

Data-center interconnections
hybrid optical switching

Interconnect Topologies
Hybrid and Optical

HyPaC

Helios

DOS

Scheme in

HyPaC Architecture

Packet Switching Network

Circuit Switching Network

Servers

C-Through

C-Through is a HyPaC network design that recruits end hosts to perform traffic monitoring and uses a partition approach to separate the circuit(optical) and packet(electrical) networks.

Advantage: buffering data in the hosts

Helios Architecture

Helios

Helios is a 2-level multi-rooted tree of pod switches and core switches. The core consists of both traditional electrical packet switches and MEMS-based optical circuit Switches.

P.S: For Modular Data Centers, A self-contained shipping container complete with servers, network and cooling, which many refer to as a Pod. Each Pod typically holds between 250 and 1000 servers.

Routing Algorithms
hybrid optical switching

The routing algorithms can affect significantly the performance of the network therefore efficient routing schemes must be deployed

in the optical networks.

In the case of the hybrid schemes (c-Through and Helios), the electrical network is based on a tree topology while the optical network is based on direct links between the nodes.

Therefore, in this case the routing is performed by a centralized scheduler that performs a bipartite graph allocation and assigns the high bandwidth requests to the optical links. If a packet has to be transmitted on a server with an established optical link, then it is forwarded directly to the optical network; otherwise it is routed through the electrical network.

In the case of the DOS architecture, the packets are sent directly to the AWGR switch and a control plane is used to route the packets by controlling the tunable wavelength converter.

The main drawback of this scheme is that the scheduler in the control plane must be fast enough to sustain the scheduling of the packets.

In all the other schemes the routing is performed at the node level, where each packet is forwarded to different port tuned at specific wavelength based on the destination address.

Routing optimization solutions for
Data Center Traffic Engineering

Equal-Cost Multi-path(ECMP)

Valiant Load Balancing(VLB)

CamCube

P.S: Traffic within a data-center network is often measured and characterized according to flows. Which are sequences of packets from a source to destination host.

Individual Flows classification

The most common classification is bimodel, using the so called Elephant and mice classes.

Elephant flows have a large number of packets and are generally long lived, they exhibit bursty behavior with a large number of packets injected in the network over a short time.

\\DROBO-FS\QuickDrops\JB\PPTX NG\Droplets\LightingOverlay.png

Edit Master text styles

Second level

Third level

Fourth level

Fifth level

\\DROBO-FS\QuickDrops\JB\PPTX NG\Droplets\LightingOverlay.pngClick to edit Master title style

Click to edit Master subtitle style



Second level

Third level

Fourth level

Fifth level





Second level

Third level

Fourth level

Fifth level


Second level

Third level

Fourth level

Fifth level




Second level

Third level

Fourth level

Fifth level



Second level

Third level

Fourth level

Fifth level


\\DROBO-FS\QuickDrops\JB\PPTX NG\Droplets\LightingOverlay.png



Second level

Third level

Fourth level

Fifth level



Click icon to add picture































Second level

Third level

Fourth level

Fifth level



Second level

Third level

Fourth level

Fifth level

Internet

Rapid Survey on Routing in Data Centers